Proc. Natl. Acad. Sci. USAVol. 91, pp. 153-157, January 1994Biochemistry

Deletion of the E4 region of the genome produces adenovirusDNA concatemersMICHAEL D. WEIDEN AND HAROLD S. GINSBERG*Departments of Medicine and Microbiology, College of Physicians and Surgeons, Columbia University, 630 West 168th Street, New York, NY 10032

Contributed by Harold S. Ginsberg, September 20, 1993

ABSTRACT Two mutants containing large deletions in theE4 region of the adenovirus genome H5dl366 (91.9-98.3 mapunits) and H2dl808 (93.0-97.1 map units) were used to inves-tigate the role of E4 genes in adenovirus DNA synthesis.Infection ofKB human epidermoid carcinoma cells with eithermutant resulted in production of large concatemers of viralDNA. Only monomer viral genome forms were produced,however, when mutants infected W162 cells, a monkey kidneycell line transformed with and expressing the E4 genes. Dif-fusible E4 gene products, therefore, complement the E4 mutantphenotype. The viral DNA concatemers produced in d1366- andd1808-infected KB cells did not have any specific orientation ofmonomer joining: the junctions consisted of head-to-head,head-to-tail, and tall-to-tail joints. The junctions were cova-lently linked molecules, but molecules were not preciselyjoined, and restriction enzyme maps revealed a heterogeneoussize distribution ofjunction fragments. A series ofmutants thatdisrupted single E4 open reading frames (ORFs) was alsostudied: none showed phenotypes similar to that of d1366 ordi808. Mutants containing defects in both ORF3 and ORF6,however, manifested the concatemer phenotype, indicatingredundancy in genes preventing concatemer formation. Thesedata suggest that the E4 ORFs 3 and 6 express functions criticalfor regulation of viral DNA replication and that concatemerintermediates may exist during adenovirus DNA synthesis.

Adenovirus DNA replication has been extensively studied(reviewed in refs. 1 and 2). Electron microscopy demon-strated that replication initiates at the ends of the lineargenome and proceeds via strand displacement, producing asemiconservatively replicated daughter genome and a single-stranded (type II) molecule which may form a duplex pan-handle by virtue of inverted terminal repeats (ITRs). Thistype II molecule then is the substrate for another round ofDNA synthesis producing a second daughter genome (3).

Insight into the mechanism of the initiation of replicationhas been gained from in vitro reconstitution ofDNA synthe-sis using isolated viral DNA and plasmid constructs. Geneticdata and an in vitro model ofDNA replication defined the roleof three essential viral proteins: a single-stranded-DNA-binding protein, a DNA polymerase, and a terminal proteinthat is covalently linked to the 3' terminal deoxycytosine ofthe genome and is critical for the initiation ofDNA synthesis(4-6). Four host proteins have also been defined as importantfor viral DNA replication: nuclear factor I and nuclear factorIII/Oct-1, which are transcription factors (ref. 7 and refer-ences therein); nuclear factor II, a type I topoisomerase (8);and pL, a 5'-to-3' exonuclease (9).An in vivo model for viral DNA synthesis used plasmid

constructs to investigate the sequence requirements for rep-lication. The ITR is essential, and a plasmid containing asingle complete copy of the repeat can be rescued by dupli-cative transfer of one end of the genome to the other (10) or

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

by replication-induced recombination (11). The last 45 bp ofthe genome contain the necessary sequence information forreplication, and any deletion of the terminal cytosine abol-ishes DNA synthesis (12, 13).Other viral gene products may be involved in viral DNA

synthesis, and other mechanisms not yet elucidated may berequired for completion of viral DNA replication. Somephenomena not accounted for in the current model includethe following: circular, supercoiled, and multimeric viralDNA molecules, which were identified in electron micro-scopic studies (14); the pL 5'-to-3' exonuclease required inthe in vitro model (9) produces defective strands that needadditional processing for maturation; a large proportion ofviral particles produced during infection carry partially rep-licated single-stranded DNA (15); and concatemers of inputDNA are formed in the in vivo plasmid model of viral DNAsynthesis (10-13). Topoisomerase, which is required in the invitro model, should not be needed to relieve tortional stressin the linear molecules predicted by the current model. Inaddition, more than the three viral genes defined by thecurrent model are needed for viral DNA replication, sincemutations in the E4 region have a significant effect on viralDNA synthesis (16-18).To gain a better understanding of the functions of the E4

genes in adenovirus DNA synthesis, the DNA replicationphenotypes of E4 mutants were investigated. These experi-ments revealed the formation of viral DNA concatemers,which suggests that multimeric genomes are a consequenceof disruption of a normal viral gene function and that themechanism ofadenovirus DNA synthesis may differ from thepresently accepted model.

MATERIALS AND METHODSCell Culture and Viruses. Monolayer cultures of KB cells

(human epidermoid carcinoma cell line) and W162 cells [aVero monkey kidney cell line which contains an integratedcopy of the adenovirus type 5 (AdS) E4 region was suppliedby G. Kettner, The Johns Hopkins University] were grownin Dulbecco's modified Eagle's medium (DMEM) supple-mented with 10%o calf serum. E4 mutants H5dl366, in351,in352, d1355, d1356, d1358, and d1359 were supplied by T.Shenk (17); H2dl808 was supplied by G. Kettner (16); and E4inORF3, E4 inORF3/inORF6, E4 inORF3/dl355, andd1366+ORF3 were supplied by P. Hearing (18). See Fig. 1 forlocation of mutations.

Purified virions, prepared by CsCl equilibrium densitycentrifugation, were used in all infections. Material isolatedfrom the gradient was disrupted with 0.1% SDS and A260 wasmeasured. An A2w of 1 indicates 1012 particles per ml (17).When purified Ad5 was titrated on KB cells, a 10:1 ratio oftotal particles to plaque-forming units was present. In all

Abbreviations: Adn, adenovirus type n; ITR, inverted terminalrepeat; ORF, open reading frame; PFGE, pulsed-field gel electro-phoresis.*To whom reprint requests should be addressed.

153

Dow

nloa

ded

by g

uest

on

June

25,

202

1

154 Biochemistry: Weiden and Ginsberg

90 92 94 96 98 100 maP units

3000 2000 1000

ORF6/7 ORF3/4

IF7 ~K~12IF6 IORF41ORF IF2 I ORF I

*{| | | { § ~~~+Sbp in351

Proc. Natl. Acad. Sci. USA 91 (1994) 155

The deletion in d1808 produces a genome that is 33 kb long(16). The even spacing and molecular size of each band arecompatible with concatemer formation in which unit ge-nomes are added to one another. As the infection progressed,the number of bands increased, and the proportion of DNAin the highest molecular weight DNA increased. These datasuggest that the lower molecular weight products are precur-sors to the higher molecular weight DNA.

Effect of Mutation of Genes in the E4 Region. To definewhich ORF in E4 was responsible for production of concate-mers, a series of mutants were used in which each E4 ORFwas individually mutated. The in351, in352, d1355, d1356,d1358, d1359, and inORF3 mutants produced only singlebands (Fig. 2B; see Fig. 1 for the ORF affected by each ofthese mutations). E4 ORF3 and ORF6 are redundant regu-lators of viral DNA synthesis (18). Therefore, a series ofdouble mutants in ORF3 and ORF6 were tested for theconcatemer phenotype. Individual mutations in d1355 (anORF6 mutation) and in inORF3 (an ORF3 mutation) synthe-sized only monomer genomes (Fig. 2B). When the doublemutants inORF3/dl355 and inORF3/inORF6 were used, aladder of bands similar to that seen with the large E4mutations was produced (Fig. 2B). Finally, when only ORF3was added back to the large E4 deletion of d1366, no con-catemers were formed (Fig. 2B, dl366+ORF3). These datademonstrate that suppression of concatemers is a combinedeffect of ORF3 and ORF6 of E4. Only monomer genomeswere produced in any mutant in which either ORF3 or ORF6remained intact.

Structure of the DNA Junctions. The finding that multimerswere synthesized during viral DNA synthesis of E4 mutantsmade defining the orientation of the genomes at the concate-mer joints important for understanding the mechanism ofconcatemer formation. A restriction map of the junctionmolecules in the multimers of d1366 and d1808 was attemptedusing the restriction enzyme HindIII. This experiment gavea surprising result. Instead of producing one or more newrestriction fragments that would define the orientation of thejunction (see below), a heterogeneous distribution of frag-ments was generated ranging in size from 2 kb to 10 kb [Fig.3A, lane 5 (d1366) and lane 8 (dI808)]. The size of the ElAterminal HindIII fragment from a monomer genome was 2.8kb in AdS, d1366, and d1808 (Fig. 3A, lanes 1, 4, and 7). TheE4 terminal HindIll DNA fragments were 1.1, 1.7, and 1.8kb, respectively (data not shown). The largest junction frag-ment predicted from ligation of the El end to itself is 5.6 kb,but when HindlIl digests ofmultimeric genomes were probed

A

with ElA DNA the smear extended beyond 10 kb, implyingthat a process other than simple ligation of one genome toanother was operating. Also, fragments smaller than thestarting 2.8 kb were produced, demonstrating that either again of new HindIII sites or a loss of genetic informationoccurred in some of the molecules generated during concate-mer formation.One possible explanation of the smear of junction frag-

ments is that the DNA molecules in the smear are highlystructured and therefore migrate aberrantly in agarose. An-other possible explanation is that the terminal fragments havelong single-stranded components that hybridize to one an-other. To test these possibilities, DNA from dl366-infectedcells was used in two-dimensional gel electrophoresis (Fig.3B). The first dimension used a nondenaturing agarose gel.The second dimension employed a denaturing alkali gelelectrophoresed with 50 mM NaOH. If the smear was dueeither to an aberrant DNA structure or to annealing of onefragment to another, the molecules present in the smearshould migrate at different molecular sizes in the first andsecond dimensions. If, however, a majority of the junctionmolecules were linear and covalently intact, then a diagonalshould be formed in which the relative migration undernondenaturing and denaturing conditions was the same. Thetwo-dimensional gel in Fig. 3B demonstrates a diagonal inwhich the apparent molecular sizes under nondenaturing anddenaturing conditions were the same. These findings indicatethat the smear of the junction fragments was predominantlycomposed of linear, covalently linked molecules.

Orientation of Concatemer Junctions. Given the imprecisejoints demonstrated in the HindIlI map, another restrictionenzyme, which produces larger fragments, may be moreuseful to map the orientation of the junction between twogenomes in the concatemer. BamHI cuts once in AdS wildtype and d1366, producing two fragments of 21.5 and 14.5 kbin wild type (Fig. 4A, lane 1) and two fragments of 21.5 and10 kb in d1366 (Fig. 4A, lane 4). The ElA probe hybridizes tothe 21.5-kb fragment of Ad5 wild type and d1366 (Fig. 4B,lanes 1 and 4), which will be called the head of the genome.The E4 probe hybridizes to the 14.5-kb fragment ofAdS wildtype (Fig. 4C, lane 1) and to the 10 kb fragment of d1366 (Fig.4C, lane 4); this E4 fragment will be termed the genome tail.To determine whether the concatemers consisted of head-

to-head and tail-to-tail genomes (i.e., El joined to El, and E4joined to E4) or head-to-tail genomes (i.e., El joined to E4)BamHI restriction maps of the junction fragments wereprepared. If the genomes of d1366 were joined head-to-head

B First Dimension -

Size 1 2 3 4in Kb+ :. ..

5 6 7 8 9

2.8

Sizee 8in Kb

C,

(D

2.8,

2.8

FIG. 3. Southern blot analysis of agarose gelelectrophoresis of HindIII-digested DNAprobed with ElA. (A) Lanes 1-3, Ad5 wild type;lanes 4-6, d1366; lanes 7-9, d1808; lanes 1, 4, and7, cesium chloride gradient-purified virus; lanes2, 5, and 8, total DNA obtained from KB cells 48hr after infection; lanes 3, 6, and 9, total DNAobtained from W162 cells 48 hr after infection.(B) Two-dimensional agarose electrophoresis ofDNA obtained from KB cells 48 hr after infec-tion with d1366. The first dimension was rununder nondenaturing conditions. The seconddimension was run under denaturing conditions.

Biochemistry: Weiden and Ginsberg

Dow

nloa

ded

by g

uest

on

June

25,

202

1

156 Biochemistry: Weiden and Ginsberg

A 1 2 3 4 5 6 7

B 1 2 3 4 5 6 7

... ..

10404 _i21b d_

C 1 2 3 4 5 6 7

4- m33

210-1SW2I3....,0 *w14_

-d21

,

FIG. 4. PFGE of adenovirus DNA. Numbers with arrowheadsindicate predicted sizes of bands in kilobases. (A) Ethidium bromidestain. Lane 1, BamHI-digested AdS wild type; lane 2, undigestedd1366 from infected KB cells; lane 3, BamHI-digested d1366 from KBcells; lane 4, BamHI-digested d1366 from W162 cells; lane 5, undi-gested d1808 from KB cells; lane 6, BamHI digested d1808 from KBcells; lane 7, BamHI digested d1808 from W162 cells. (B) Southernblot of the same gel shown inA probed with ElA DNA. (C) Southernblot of the same gel probed with E4 DNA.

in the concatemers, the junction created by multimerizationshould be -42 kb and would hybridize only with an ElAprobe. If the joints were head-to-tail, then the newly createdfragment would be 32 kb and would hybridize with both ElAand E4 DNAs. If the joints were tail-to-tail, then the newlycreated fragment would be 20 kb and would hybridize withonly E4 DNA.When BamHI-digested d1366 concatemers were examined,

three new bands were formed. Two additional bands of 32and 42 kb were seen in the ethidium bromide-stained gel afterPFGE (Fig. 4A, lane 3), and a third new band, which was notresolved from the 21.5-kb head fragment on ethidium bro-mide staining, was apparent at 21 kb when a Southern blot ofthe pulsed-field gel was probed with E4 DNA (Fig. 4C, lane3). The new 42-kb band hybridized only with ElA DNA (Fig.4B, lane 3) and therefore represents.a head-to-head junctionmolecule. The new 32-kb band comigrated with the monomerband of uncut d1366 (compare the bands marked 32 in Fig. 4,lanes 2 and 3) which hybridized with both ElA DNA (Fig. 4B,lane 3) and E4 DNA (Fig. 4C, lane 3). This monomer-lengthband, therefore, represents a head-to-tail junction molecule.The predicted size of the tail-to-tail junction is 20 kb. Noadditional band was apparent at this size in the ethidiumbromide-stained gel, but a new band was seen with the E4probe (Fig. 4C, lane 3, marked 21). Therefore, the tail-to-tailjunction fragment existed but was not resolved from the headflanking fragment in this gel system. These data indicate thatall possible junction molecules existed in d1366 concatemersand that the joining reaction was not orientation-specific.The interpretation of the d1808 PFGE banding pattern was

more difficult, since BamHI cuts this Ad2 derivative at twosites (the site at base pair 15403 is missing; L. H. Epstein and

C. H. Young, personal communication), yielding two 10-kbbands and one 11-kb band. Therefore, the three newjunctionmolecules would range in size from 20 to 22 kb. Although asmear of new fragments representing junction molecules wasseen around 21 kb when concatemers of d1808 were digestedwith BamHI (Fig. 4, lane 6), the individual forms were notseparated from each other by PFGE.

Therefore, a cloning approach was undertaken to deter-mine the types ofjunction molecules present in this mutant.Two DNA libraries were made from HindIII-digested d1808DNA and Ad5 wild-type DNA. The viruses were isolatedfrom infected KB cells, and the digested DNA was insertedinto the HindIll site of pUC18. When the Ad5 wild-typelibrary was probed, no ElA-hybridizing colonies were ob-tained. Terminal HindIII restriction fragments of Ad5 wildtype should not be represented in this library, since one endof the terminal fragment is blunt. The DNA library derivedfrom d1808 had 16 colonies that hybridized with ElA and/orE4 DNA. Four colonies hybridized only with ElA DNA and,therefore, represented head-to-head junctions; 8 colonieshybridized only with E4 DNA and thus represented tail-to-tail junctions; and 4 colonies hybridized to both ElA and E4probes and represented head-to-tail junctions. Therefore, allpossible joint orientations existed in the viral concatemers ofd1808-infected as well as d1366-infected KB cells.

Complementation of E4 Deletion Mutants. The concatemersthat were produced with the E4 deletion mutants could be dueto loss of gene function encoded in the E4 region or the resultof some structural feature of the mutated viral DNA thatallowed annealing of one genome to another. The presence ofa copy of the E4 region incorporated into the chromosome ofthe W162 cell line allowed testing of this question. WhenDNA was isolated from d1366- or d1808-infected W162 cells,only monomer viral DNA was detectable (data not shown).This demonstrated complementation of the E4 deletion phe-notype by E4 gene expression. Data supporting this conclu-sion are shown in Figs. 3 and 4. When d1366 or d1808 waspropagated in KB cells and the DNA was digested withBamHI (Fig. 4A, lanes 3 and 6) or HindIII (Fig. 3A, lanes 5and 8), bands not predicted from the restriction map of themonomer were present. These new bands represent junctionmolecules (see above). When these mutants were used toinfect W162 cells, the only bands that appeared were pre-dicted from the monomer restriction map (Fig. 4A, lanes 4and 7; Fig. 3A, lanes 6 and 9). This finding demonstrates thatthe mutant phenotype could be complemented by expressionof E4 genes that were not physically linked to the viralgenome. These data indicate that the products of the E4ORF3 and ORF6 are essential for normal viral DNA synthe-sis and, hence, responsible for prevention of concatemeraccumulation.

DISCUSSIONMutations that affect both ORF3 and ORF6 of E4 were foundto form DNA concatemers during viral synthesis in KB cells.The concatemers were formed by covalent joints that werenot orientation-specific and frequently involved loss or gainof DNA sequences. Concatemers were not observed whenmutants infected W162 cells, a line transformed with andexpressing E4 genes, which demonstrated that diffusible E4gene products complemented the E4 mutations. There aretwo possible explanations of this observation. (i) The con-catemer-forming mutants have lost gene functions (ORF3 andORF6) that alter the activity of mammalian factors, andwithout the inhibition or stimulation of these host factors,normal viral replication intermediates are drawn into a dead-end pathway in which concatemers are formed. (ii) ORF3 andORF6 are directly involved in the pathway of viral DNAreplication, and concatemer accumulation occurs when the

Proc. Natl. Acad Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

June

25,

202

1

Proc. Natl. Acad. Sci. USA 91 (1994) 157

pathway is blocked. These two sets of possibilities are notmutually exclusive.

Host-mediated ligation has been invoked to explain theappearance of concatemers of input plasmids containingadenovirus origins of replication that was observed in the invivo model of viral DNA replication (10, 12, 13). Thisexplanation of plasmid concatemer formation is an exampleof a dead-end pathway in which concatemers are formedbecause the host DNA ligase produces multimeric genomes.Activity of a mammalian ligase is not, however, sufficient toexplain concatemer formation of test plasmids, becausereplicating constructs in the in vivo model have covalentlyattached terminal protein (10) and, therefore, would not besubstrates for ligase until the serine adduct had been removedfrom the 5'-terminal phosphate. At least one other enzymaticactivity, such as a cellular endonuclease or exonuclease,would be required to remove the protein adduct. Alterna-tively, a viral molecule could be produced by a cellular DNAprimase-polymerase that could initiate viral DNA replicationwithout a terminal protein. An exonuclease, endonuclease,or DNA primase-polymerase would create a viral genomethat had lost DNA sequences and was a substrate for DNAligase. Another possibility is that small regions of homologypresent in direct and inverted subrepeats contained in theITR of the adenovirus genome could be a substrate for a hostrecombination pathway. Recombination in a palindromewithin an ITR of two different viral genomes would producemultimers directly. If recombination occurred at a low fidel-ity with internal subrepeats in the ITR, DNA sequences couldbe lost. Any of these four enzyme activities could lead tojunctions with imprecise joints and no orientation specificity,which were observed in d1808 and d1366.The mechanism by which the E4 mutants could cause

persistence of host factors responsible for concatemer for-mation may be through loss of the host shutoffphenotype. Innormal viral infection, host DNA synthesis and proteinsynthesis are shut off (17). With mutants in which E4 ORF6is the only E4 gene expressed, host proteins are efficientlyshut off, and viral replication is normal (21). But afterinfection with mutants in ORF6, host DNA and proteinsynthesis persist 24 hr into infection (16, 18). ORF6 mutantswould be expected to have higher levels of host enzymeactivity than would be found in wild-type adenovirus infec-tion. The abnormal persistence of host enzymatic activitiessuch as an endonuclease, recombinase, or DNA polymeraseis not sufficient to explain the concatemer phenotype, sincemutants such as d1355, an ORF6 mutant that does notsuppress host DNA synthesis, form only monomers.

Alternatively, concatemers could be normal intermediatesof viral DNA synthesis, and ORF3 and ORF6 could beinvolved in the pathway that converts multimers to mono-mers. Evidence that multimers exist in adenovirus replicationcomes from electron microscopy (14). When protease is notused in viral DNA isolation, multimers and circular mole-cules are frequently found; protease treatment resolves thesemolecules to unit genomes (14). The electron microscopicstudy that defined the currently held model for viral DNAreplication did use protease for viral DNA preparation (3).Other data that support the possibility that multimers areformed during replication come from the in vivo model ofadenovirus DNA replication in which multimers are formed(10-13).A rolling circle model found in bacteriophage OX174 or a

rolling hairpin model found in adeno-associated virus wouldproduce concatemers (22), but the former produces onlyhead-to-tail molecules and the latter produces only head-to-head molecules that progress from one to two to four to eight

genome lengths. The finding of head-to-head and head-to-tailjunctions with size increments ofone genome length in the E4mutants suggests that both models of adenovirus DNA syn-thesis are unlikely.One model that could account for the current observations

is a replication-associated overlap recombination in whichthe leading-strand synthesis is protein-primed from both endsof the genome and produces single-stranded molecules thatare available for duplex formation and completion of thegenome. Concatemers would be formed after priming of apalindrome of one ITR with the reiterated sequences ofanother ITR. The occurrence of the palindrome within theITR would allow head-to-head and head-to-tail molecules toform. Such overlap recombination was observed betweentwo plasmid constructs in which each carried a single copy ofthe adenovirus ITR (11). The T7 bacteriophage presentsanother model in which linear DNA molecules that carryterminal redundancies replicate through concatemer inter-mediates (22). The 5'-to-3' exonuclease needed in the in vitromodel of adenovirus DNA replication (9) would producemolecules similar to those of T7. Neither of these modelspredicts the loss of DNA sequences at the junctions asobserved with H5dl366 and H2dl808 DNA synthesis. If amultimer is in the pathway of viral DNA replication, thenORF3 and ORF6 would be involved both in the fidelity ofthereaction and in resolving higher-order forms to monomers.

We are most grateful to Drs. Thomas Shenk and Gary Kettner fortheir generous supply of viral mutants. M.D.W. is a Parker B.Francis Fellow in pulmonary research. This research was supportedby Grant AI12052 from the National Institute of Allergy and Infec-tious Diseases.

1. Kelly, T. J., Wold, M. S. & Li, J. (1988) Adv. Virus Res. 34,1-42.

2. Stillman, B. (1989) Annu. Rev. Cell Biol. 5, 197-245.3. Lechner, R. L. & Kelly, T. J. (1977) Cell 12, 1007-1020.4. Challberg, M. D., Desiderio, S. V. & Kelly, T. J. (1980) Proc.

Natl. Acad. Sci. USA 77, 5105-5109.5. Ikeda, J., Enomoto, T. & Hurwitz, J. (1981) Proc. Natl. Acad.

Sci. USA 78, 884-888.6. Stillman, B. W. (1981) J. Virol. 37, 139-147.7. Mul, Y. M., Verrijzer, P. & van der Vliet, P. C. (1990) J. Virol.

64, 5510-5518.8. Nagata, K., Guggenheimer, R. A. & Hurwitz, J. (1983) Proc.

Natl. Acad. Sci. USA 80, 4266-4270.9. Kenny, M. K., Balogh, L. A. & Hurwitz, J. (1988) J. Biol.

Chem. 263, 9801-9808.10. Hay, R. T., Stow, N. D. & McDougall, I. M. (1984) J. Mol.

Biol. 175, 493-510.11. Ahern, K. G., Wang, K., Xu, F., Mathews, C. Z. & Pearson,

G. D. (1991) Proc. Natl. Acad. Sci. USA 88, 105-109.12. Hay, R. T. (1985) EMBO J. 4, 421-426.13. Bernstein, J. A., Porter, J. M. & Challberg, M. D. (1986) Mol.

Cell. Biol. 6, 2115-2124.14. Robinson, A. J., Younghusband, H. B. & Bellett, A. J. D.

(1973) Virology 56, 54-69.15. Daniell, E. (1976) J. Virol. 19, 685-708.16. Weinberg, D. H. & Kettner, G. (1986) J. Virol. 57, 833-838.17. Halbert, D. N., Cutt, J. R. & Shenk, T. (1985) J. Virol. 56,

250-257.18. Huang, M. & Hearing, P. (1989) J. Virol. 63, 2605-2615.19. Van der Ploeg, L. H. T., Schwartz, D. C., Cantor, C. R. &

Borst, P. (1984) Cell 37, 77-84.20. Weiden, M., Osheim, Y. N., Beyer, A. L. & Van der Ploeg,

L. H. T. (1991) Mol. Cell. Biol. 11, 3823-3834.21. Hemstrom, C., Virtanen, A., Bridge, E., Kettner, G. & Pet-

tersson, U. (1991) J. Virol. 63, 1440-1449.22. Kornberg, A. & Baker, T. (1992) DNA Replication (Freeman,

New York), pp. 298-303, 592-597, 700-703.

Biochemistry: Weiden and Ginsberg

Dow

nloa

ded

by g

uest

on

June

25,

202

1